When slicing yoda.stl it reduces peak memory usage by 5% and speeds up slicing by 13%. Further optimization using a simple proxy to implement the two methods (size/extends) needed on the merge meshes is also possible and undoubtedly worthwhile.